High Energy Phenomena in Massive Stars
ASP Conference Series, Vol. XX, 2010
Josep Martı́ et al.
Non-thermal radio emission from OB stars: an observer’s
view
arXiv:0904.0533v1 [astro-ph.SR] 3 Apr 2009
Paula Benaglia1,2
(1) Instituto Argentino de Radioastronomı́a, CCT La Plata-CONICET,
C.C. 5, 1894 Villa Elisa, Argentina
(2) Facultad de Ciencias Astronómicas y Geofı́sicas, Paseo del Bosque
s/n, 1900 La Plata, Argentina; pbenaglia@fcaglp.unlp.edu.ar
Abstract.
Some early-type stars are detectable radio emitters; their spectra
can present both thermal and non-thermal contributions. Here I review the
public radio data on OB stars, focusing on the non-thermal sources. The analysis
of the statistical results gives rise to many open questions that are expected to
be addressed, at least in part, with the upgrades of current radio telescopes and
the upcoming new generation instruments.
1.
Introduction
Early-type stars like OB stars and their descendants Wolf-Rayet (WR) stars are
characterized by strong winds that continuously expel matter from the stellar
atmospheres. The velocity of the outflow can reach thousands of km s−1 , and
the mass is lost at rates up to 10−4 M yr−1 . The plasma forming the winds is
optically thick and radiates at radio continuum by thermal Bremsstrahlung. The
winds are prone to also suffer instabilities that give rise to shocks. In massive
binary systems, regions where winds from the two stars collide are permeated
by even stronger and larger shocks, which, in turn, accelerate particles up to
relativistic energies, and non-thermal (synchrotron) emission is generated.
In the case of uniform, homogeneous winds, the measurement of the thermal
radio-flux density at a given frequency allows the estimate of an average value of
the stellar mass loss rate, provided some basic wind parameters are known, and
the stellar distance is determined. The result will be an overestimation if there
are inhomogeneities in the plasma or if the measured flux density has contribution of thermal and non-thermal emission. The velocity at which a massive star
loses mass is a fundamental input in stellar evolution and population models,
and slightly different values can change the results drastically.
Radio observations at more than one frequency - preferentially, along the
spectrum - are important to describe the emission regime, to disentangle the
different contributions to the flux, and to study the emission mechanisms that
lead to the observed spectrum. The detection of synchrotron emission is by itself
an evidence for the existence of relativistic particles. These particles, by interactions with other ones and with magnetic and photon fields are also involved
1
2
P. Benaglia
in the production of high-energy radiation1 . Thus, the study of the radio emission from the stellar winds can give information on the population of relativistic
leptons in the winds, and about the possible non-thermal high energy emission
from massive stars (De Becker 2007). Moreover, polarization measurements at
non-thermal stellar sources can provide insights on the magnetic field of massive
stars, one of the stellar parameters most difficult to measure.
In the last decades, hundreds of OB stars have been observed, many of them
at more than one wavelengths. A study of the detected cases is presented here,
based mainly on a statistical analysis. Examples of instrumental developments,
in either existent or new facilities, that will easily allow to enlarge the number
of detections by more than an order of magnitude closes this review.
2.
Milestones in stellar wind studies
It could be said that the history of the studies on stellar winds begun with
the discovery of SNe 3c10, the “Tycho” supernova, in 1572, and continued with
the monitoring of subsequent novae, supernovae, and LBVs magnitude changes.
With time, as spectroscopy became a powerful tool, C. Wolf and G. Rayet (1867)
discovered that a few stars showed unusually broad emission line spectra. PCygni profiles were measured (Campbell 1892), and they were interpreted by
taking into account Doppler shifts, revealing signatures of outflows.
It was early in the XXth Century that the hypothesis that selective absorption can drive the outflows was formulated (Milne 1924, Johnson 1925). L.
Biermann, among others, proposed the existence of a continuum solar ejecta
and expanding envelopes, and also discovered the first evidence of that, through
the deflection of a comet tail (Biermann 1951). Surprisingly, it was not until
1958-60, that the expressions ‘solar wind’ and ‘stellar wind’ were coined (Parker
1958, 1960). By those years, the OB-winds studies greatly benefited from the
progress and knowledge that were being gained on the solar wind. In the year
1967 the Mariner 2 probe made the first detection of the solar wind. In the
same year the first OB-UV spectra were measured by D. C. Morton using a detector on a rocket. Breakthroughs began to accumulate in the seventies: Lucy
& Solomon (1970) developed a radiation-driven wind model, radio observations
toward massive stars were started, the CAK (Castor, Abbott & Klein 1975)
theory burst in and good predictions on mass loss rates and terminal velocities
became available. Wright & Barlow (1975) and Panagia & Felli (1975) proposed
a thermal description for the emission of stellar winds. The first image of a
resolved wind was published in 1989 (Moran et al. 1989), and soon VLBI mapping, phase monitoring, and multifrequency studies on stellar winds begun.
1
A key point is whether the associated high-energy radiation will be detected by the present
telescopes.
Non-thermal OB stars
3.
3
Stellar wind regions for particle acceleration
In a system composed by two stars with winds there are some regions with
shocks that are capable to accelerate particles up to very high energies, namely,
the wind of each star is prone to suffer instabilities that produce shocks, the region where single supersonic winds encounters the interstellar matter, the region
where the two winds collide (see Figure 1). The latter region has a rather strong
magnetic field, related to the stars. In such a scenario non-thermal emission can
be produced. The same population of relativistic particles will be involved in
processes that also produce high-energy photons. Electrons accelerated in the
electrostatic field of the ions give rise to relativistic Bremsstrahlung emission.
The copious UV photons from the OB stars are boosted by relativistic electrons
to higher frequencies (IC scattering). Synchrotron emission is produced by electrons accelerated at the existent magnetic field. Neutral pions generated by
inelastic collisions between relativistic and thermal protons decay into gamma
rays. The possibility that high energy emission can also be produced at the environs of early-type stars motivates the search for counterparts to the unidentified
gamma-ray sources, e.g. Benaglia et al. (2005), Romero et al. (1999).
Figure 1.
Possible regions for shock-mediated particle acceleration in massive star winds: region a) where, in a massive binary system, two stellar
winds collide; region b) in the unstable single stellar wind of an early-type
star; region c) at the terminal shock produced when and where the stellar
wind encounters the ISM.
4.
Published radio continuum data on OB stars
In 1995, a catalogue of all observed stars at radio wavelengths till that date was
presented by Wendker (1995). It listed information on 3021 stellar objects with
spectral types from O to M. About 275 were cataloged as O − B2 stars, of which
around 40 were detected, at one or more radio frequencies. The works cited by
Wendker on the detected cases were reviewed in order to derive spectral index
values. The bibliographic sources consulted were Gibson & Hjellming (1974),
4
P. Benaglia
Abbott et al. (1980, 1984), Drake et al. (1987), Persi et al. (1988), Bieging
et al. (1989), Drake (1990), Leitherer & Robert (1991), and Howarth & Brown
(1991), Dougherty (1993).
A search of OB stars radio observations in the literature since 1995 to date
yields the papers by Leitherer et al. (1995), Contreras et al. (1997), Scuderi et
al. (1998), Benaglia et al. (2001), Rauw et al. (2002), Blomme et al. (2002,
2003), Setia Gunawan et al. (2003a, 2003b), Benaglia & Koribalski (2004), De
Becker et al. (2004, 2005), Blomme et al. (2005), Puls et al. (2006), Benaglia
et al. (2006), Blomme et al. (2007), Benaglia & Koribalski (2007), Benaglia et
al. (2007), Schnerr et al. (2007), Petr-Gotzens & Massi (2007), van Loo et al.
(2008), and Dougherty & Kennedy (2009).
Table 1 lists the stars detected at least at one frequency, their spectral type,
and references. A spectral index was derive whenever possible. The listed stars
are divided in three groups according to the spectral indices, labeled as ‘Uncl’,
‘NT’, and ‘T’. When a spectral index α is > +0.3 (Sν ∝ ν α ) the source is classified as ‘Th’ (thermal); if α → 0.0 or less, the source is ‘NT’ (non-thermal).
For stars detected at only one frequency, the spectral index remains unknown,
or unclassified (‘Uncl’). The third column quotes the reference from where the
radio data were gathered. It is usually the most recent reference, which refers to
previous ones. It does not necessarily mean that the observations were carried
out by the authors of the paper. An entry (NT) means “probably NT”. It is
classified in this way because either it is suggested at the reference quoted, or
it was detected at one frequency, and the detection limit at a longer frequency
is lower than the flux detected at the shorter frequency. A note (Th) means
“probably thermal”, as proposed by the corresponding reference, or it was detected at one frequency, and the detection limit at a shorter frequency is lower
than the flux detected at the longer one.
The spectral types were mostly taken from the GOS Catalogue (Maı́zApellániz et al. 2004), except when otherwise notice. Spectral information
on binarity, from Mason et al. (2009), is given in brackets.
4.1.
Bias and error sources
To derive the spectral index information for each case, the flux density of the
original references were used. Because of the extremely high sensitivity needed
for measuring continuum emission from stellar winds (below the mJy), and the
angular resolution needed to resolve the systems (below the arcsec), almost all
detections of OB winds revealed point sources. In this respect, the information presented here must be taken with caution. The observing campaigns were
carried out using different telescopes and even the same telescope was used at different epochs, i.e. different performances and likely different states of the source.
Table 1 gathers radio data from 26 studies (see references at Table 1, Column
3). Each group has been selected with different criterion, such as stellar distance
and luminosity, celestial regions, spectral type range, declination coverage, etc.
Different sensitivities led to fix different noise thresholds, which implies different detection limits. The various angular resolutions used yielded to resolved
Non-thermal OB stars
Table 1.
5
O – B2 stars with detected radio emission from winds.
Name
Sp.Class
Ref.
Observed ν
[GHz]
HD 93129A
HD 150136
HD 93250
HD 66811
HD 190429A
HD 16691
HD 15570
HD 164794
CD-47 4551
HD 14947
Cyg OB2 #7
Cyg OB2 #9
HD 15558
Cyg OB2 #11
HD 168112
HD 108
HD 210839
Cyg OB2 #8A
HD 124314
HD 206267
HD 167971
HD 150135
HD 152623
HD 47839
Cyg OB2 335
HD 166734
Cyg OB2#5
HD 24912
HD 47129
HD 151804
HD 19820
HD 37043
HD 57061
HD 149404
HD 76341
HD 37742
HD 149757
HD 37468
HD 36486
HD 209975
Cyg OB2 #10
HD 30614
HD 195592
HD 152424
HD 163181
MWC 349
HD 37128
HD 204172
HD 154090
HD 5394
θ Ori
HD 193237
HD 190603
HD 194279
BD-14 5037
HD 148379
HD 37017
HD 2905
HD 152236
HD 169454
HD 36485
HD 41117
HD 80077
HD 37479
O2If*+O3.5V
O3.5If*+O6V+..
O3.5V((f+))
O4 I(n)f
O4 If+
O4I f+
O4If+
O4((f))+?
O5If
O5If+
O5If+
O5If+
O5III(f)+O7V
O5III(f)
O5.5III(f*)
O6:f?pe
O6I(n)fp
O6Ib(n)(f)+O5.5III(f)
O6V(n)((f))
O6.5V((f))+O9.5:V
O6.5V+O5-8V
O6.5V((f))
O7V(n)((f))
O7V((f))+O9.5V
O7V
O7Ib(f)
O7Ia+Ofpe/WN9
O7.5III(n)((f))
O7.5I+O6I
O8Iaf
O8.5III((n))+B0V
O9III+B7IV
O9II+B2V
O9Ia+O6.5III
O9 Ib
O9.7Ib+B2III
O9.5Vnn
O9.5V+B0V+B2..
O9.5II+B0III
O9.5Ib
O9.5Ib
O9.5Ia
O9.5Ia
CO9.7Ia
BN0.5Iap
B[e]+B0III
B0 Ia
B0 Ib
B0.7 Ia
B0IVe
B0.5V+TT+*
B1 Ia
B1.5 Iae
B1.5 Ia
B1.5 Ia
B1.5 Iape
B1.5V
B1Ia
B1Ia
B1Ia
B2 V
B2Ia
B2Ia+
B2Vp
a
a
b
c
d
e
f
g
a
d
h
i
j
j
k
e
l
h
a
m
n
o
p
l,m
h
j
q
l
e
j
r
s
s
s
v
s,t
s
m
s
u
u
d,l
d
s
o
h
w
v
v
x
z
d
d
d
d
v
y
d
p
j
y
d
b
y
1.4 – 24.5
1.4,2.4,4.8,8.6
4.8,8.6
1.4,4.8,8.5,15
8.5
4.9
4.8,8.4,1.3mm
1.4,4.8,15
1.4,2.4,4.8,8.6
4.9,8.5
1.4,2.4,4.8,8.6
1.4,4.8,8.4,15
4.8
1.4,4.8
4.8,8.5,15
4.9
1.4,8.4
.4,2.4,4.8,8.6
1.4,2.4,4.8,8.6
4.9
1.4,4.9,8.5
4.8,8.6
1.4,2.4,4.8,8.6
1.4,4.9,8.5
1.4,2.4,4.8,8.6
4.9
4.9,8.5,22
1.4,8.4
4.9
4.9,15
2.7,8
8.4
8.4
8.4
8.4
4.9,8.4
8.4
4.9,8.5,15
8.4
15
15
1.4,4.9,8.5,15
4.9,8.5,15
8.4
4.8,8.6
1.4,2.4,4.8,8.6
1.4,15,20
8.5
8.5
8.5
4.9,9.5
1.4,5,8.5,15
4.9,8.5,15
8.5,15
4.9,8.5,15
17
8.5
8.5
4.8,8.6
4.9,15
8.5
4.9,8.5,15
4.8,8.6
8.5
Sp. Index
group
d
[kpc]
NT
NT
(NT)
Th
Uncl
(Th)
(Th)
NT
NT
(Th)
Uncl
NT
(NT)
(NT)
NT
(Th)
Th
NT
NT
Uncl
NT
Th
NT
NT
NT
(Th)
NT
Th
(Th)
NT
(NT)
Uncl
Uncl
Uncl
Uncl
Th
Uncl
NT
(NT)
Uncl
Uncl
Th
Th
Uncl
Th
Th
Th
Uncl
Uncl
Th
NT
Th
NT
Th
Th
Uncl
(NT)
Uncl
Th
Th
(NT)
Th
Th
(NT)
2.5
1.4
2.2
0.4
1.7
2.2
1.6
1.7
2
1.7
1.7
2,2
1.7
2
2.5
0.8
1.7
1
0.8
2
1.4
1.9
0.8
1.7
1.7
0.5
1.5
1.9
1
0.5
1.5
1.4
1.8
0.5
0.2
0.5
0.5
0.8
1.7
1
1.3
1.9
1.4
1.2
0.5
3
1.1
0.25
Status
Bin
Mult,[SB10]
[C]
[C]
[SB2?]
[C]
Cte
Bin,[SB2?]
Bin,[SB20]
[SB1?]
Bin,[SB1?]
[C]
Cte
Bin
[SB1?]
Bin
Mult,[SBE]
†
[SB1O]
Bin,[SB10]
Mult,[SB2OE]
Cte
Cte
Bin
Cte
Cte
Mult,[SB2?]
Bin,[SB10E]
Cte
SB1?
SB1?
EB††
Bin
Mult
2
2
1
1.7
1.3
SB
1.1
1.8
0.9
Bin
1.5
3
0.5
Mult
a: Benaglia et al. 2006; b: Leitherer et al. 1995; c: Blomme et al. 2003; d: Scuderi et
al. 1998; e: Persi et al. 1988; f: Lamers & Leitherer 1993; g: Rauw et al. 2002; h: Setia
Gunawan et al. 2003a; i: van Loo et al. 2008; j: Bieging et al. 1989; k: Blomme et
al. 2006; l: Schnerr et al. 2007; m: Drake 1990; n: Blomme et al. 2007; o: Benaglia et
al. 2001; p: Setia Gunawan et al. 2003b; q: Dougherty & Kennedy 2009; r: Gibson &
Hjellming 1974; s: Howarth & Brown 1991; t: Abbott et al. 1980; u: Puls et al. 2006;
v: Benaglia et al. 2007; w: Blomme et al. 2002; x: Dougherty 1993; y: Drake et al.
1987; z: Petr-Gotzens & Massi (2007); †: Barbá (private communication); †† Bulut &
Demircan 2007.
6
P. Benaglia
multiple systems in some cases, and in others a multiple system was considered
as a point source; the exact system composition, in many cases, is still unknown.
For the studies of binary systems, the epoch of the observation is often
critic, since the radio flux is expected to vary along the orbital phase. This issue
could not be taken yet into account when observing OB stars in the radio domain, since radio telescopes with still better sensitivity than that now available
and a larger uv coverage (especially for southern stars) are needed.
Some spectral indices were derived from data taken at different epochs,
months or even years apart. This is a potential source of error particularly for
massive binary systems where the flux density can be sensitive to the orbital
phase. If the system is wide (period: ∼ decades) and is away from periastron,
the spectral index derived can be taken as a reasonable average. But this is not
the case if the system is close or approaching periastron. Even if the observation
was set at two separate frequencies simultaneously, it should be considered that
the spectral index is valid only at one particular orbital phase. For systems
with short periods (hours) a 12-h synthesis observations allow only to derive a
spectral index averaged over the whole period.
An additional problem introduced when observing near-to-periastron systems is that the non-thermal emission can be absorbed by the thermal one from
the single winds, hiding a possible NT case.
The observations analyzed in the present paper were taken along 35 years:
spectral classification of the sources have varied with time, in accordance with
the determination of more accurate optical spectra. All stellar parameters derived from the measured flux densities (like the mass loss rate, for instance) are
only indicative and can be superseded by further studies.
5.
Statistical analysis and results
As mentioned in the previous Section, the thorough work by Wendker (1995)
comprised the results of observations toward 3021 stellar-like sources, 274 of
which were catalogued as O-B2 spectral type stars. It can be appreciated than
around 440 objects were detected, and about 40 of them were O-B2 stars.
If the observations taken from 1995 to 2008 are added to those recorded
by Wendker (1995), a total number of ∼ 65 O-B2 stars have been detected up
to date. Around a 25% of them lacks of information on the radiation regime;
for the rest it has been possible to derive spectral indices. Approximately half
of the α values are close to +0.6, whereas the other half are around or below
zero. Figure 2 shows the number of detected stars vs spectral type, and also the
information on the spectral index.
Non-thermal OB stars
7
Figure 2.
Number of detected O–B2 stars as a function of spectral type.
Blue bars: stars with spectral indices α equal to 0 or negative (NT). Red
bars: stars with α near +0.6 (Th). Gray bars: stars with no spectral index
information (Uncl).
Figure 3.
Percentage of detected stars with spectral index flat or negative
[NT and (NT) in Table 1], as a function of spectral type. The lower values
obtained for spectral types O4 and O9 are discussed in the text.
Besides the issues enumerated in Sect. 4.1, there is a very strong constraint
when performing a statistical analysis from the data of Table 1: the sample is
still very small. Consequently, the results that can be drawn from the analysis
must be taken as indicative in the low-significance limit. The results, rather
than providing certainties, open many questions.
Figure 3 presents the percentage of NT stars vs spectral type. It is noteworthy that out of five O4 stars, the one classified as NT is the only known to
be a binary. Why all O4 “single” (i.e. not yet known as in a binary or multiple
system) are thermal emitters? Four O9 stars have been detected, though at only
one frequency: is it a coincidence that no non-thermal O9 have been detected?
Figure 4 displays information on the ‘f’ sub-classification, characteristic of
evolved O prior to the WR stage. About 22 stars with spectral types O2 –
8
P. Benaglia
Figure 4.
Gray bars: percentage of f-tagged (detected) stars; green bars:
percentage of NT f-tagged stars, both as a function of spectral type.
Figure 5.
Percentage of NT stars that have a (secondary) companion, of any
spectral type (in purple), compared with the ones that belong to an O+OB
system (hatched purple).
O6 were detected, all tagged with ‘f’ (see Maı́z Apellániz et al. 2004 for a
comprehensive classification of Of stars). Why none no-Of star up to O6 has
been detected? Altogether, 25 stars are classified as f’: eight as ((f)), two as
(f) and 15 as f. Does this say something about the lifetime of each ‘f’ sub-phase?
Information on the binarity status of the detected stars is presented in Fig.
5. Massive secondary companions have been identified in a couple of cases by visual observations (e.g. HD 93129A). Optical spectroscopic analysis and speckle
interferometry permitted to find the spectral class of the secondary, or at least
the type of system (eclipsing, spectroscopic, etc., see Mason et al. 2009). Are
all NT emitters in massive binary systems? How is it possible to establish the
existence of ONE single NT emitter (e.g. HD 168112)? A detailed study on
binarity (visual, spectroscopic, eclipsing types) should be performed at least on
Non-thermal OB stars
9
Figure 6.
Location of the detected non-thermal OB stars (green stars) at
the galactic plane (spiral arm pattern and locations of HII regions -black open
circles- from Russeil 2003). The yellow circle represents the Sun.
the detected sample.
The spectral index derived from the observations ranges within the interval
(−2.4, 1.3): is there any consistent model to explain such a broad spectral index
range? How accurate are the error estimates in flux densities? Are we able
to state that if the mass loss rate derived from observations is larger than the
expected one (dM/dtobs > dM/dtexp ) then there is contribution of non thermal
emission?
6.
Prospects in the light of forthcoming observational facilities
The detected OB stars at radio waves are more than 60. Unfortunately, this
is a very low number when studying galactic OB stars. Moreover, there is a
flux limit on detection, strongly affected by the stellar distance. Figure 6 shows
the distribution of the sources labeled as NT and (NT) in Table 1, on the local
galactic environment: no pattern can be observed. These sources are closer than
3 kpc. Currently there is a limiting factor for detection at larger distances, and
this is the telescope observing time and achievable minimum noise. Radio telescopes with better sensitivity and angular resolution are fundamental to build
a more homogeneous sample.
10
P. Benaglia
The Expanded Very Large Array, and the e-Merlin2 upgrade will be operational soon, in 2009. In the southern hemisphere, the Australia Telescope Compact Array has very recently been updated with new receivers: the CABB (or
Compact Array Broadband Backend, www.narrabri.atnf.csiro.au/observing/).
Total bandwidths of 128 MHz are replaced with 2 GHz ones, improving sensitivity in a factor 4. The new system is available for use from the June 2009
semester.
In the near future powerful instruments will be working along the whole
radio spectrum: from a few MHz (cm) to THz (sub-mm) emission will be
probed. The Australia Square Kilometre Array Pathfinder (ASKAP) is a 36
12m-antenna interferometer with a 30 deg2 field-of-view, at present being built in
Western Australia. This instrument will observe, in principle, up to 2 GHz, with
an angular resolution of some arcsecs, and is supposed to be fully operational in
2013. The Square Kilometre Array (SKA) is planned with about 180 antennas,
observing frequencies from 0.1 to 25 GHz, beginning at the end of next decade.
These telescopes will be able to discover large numbers of NT stellar systems.
The Atacama Large Millimeter Array (ALMA) will be a mm+sub-mm array facility, built by a consortium from Europe, USA, Japan, and Canada, among
other partners. It will include 50 dishes of 12 m plus a central, more compact,
array (12 7m-antennas and 4 12m-antennas). The observing frequencies will
cover from 30 to 720 GHz. The expected angular resolution will be that of the
Next Generation Space Telescopes (JWST, SKA, ELT), and will reach tenth of
arcsecs. The instrument is due to 2012. Contributions by ALMA to the study of
early-type stars could be the detection of (thermal) fluxes of massive stars, and a
meticulous study about mass loss rates; the mapping of the wind structure with
unprecedented high dynamic range and angular resolution of 0.1”; the imaging
of the gas kinematics and distribution in stellar environments, and the study of
formation regions of massive protostars. All this conforms a promising future.
Acknowledgments. P.B. greatly acknowledges the invitation from the SOC
of the HEPIMS Workshop, and all the support including funding provided by
the LOC. Many thanks especially to Dr. Josep Martı́ and Dr. Pedro L. Luque
Escamilla. P.B. is indebted to Gustavo E. Romero for his help in all steps involved in the preparation of the present review, and wants to thank also Josep
Marı́a Paredes and Sean M. Dougherty. This work was partially supported by
the Argentine agencies CONICET (PIP 452 5375) and ANPCyT (PICT-200700848).
References
Abbott, D. C., Bieging, J. H., Churchwell, E., & Cassinelli, J.P. 1980, ApJ, 238, 196
Abbott, D. C., Bieging, J. H., & Churchwell, E. 1984, ApJ, 280, 671
Benaglia, P., Cappa, C. E., & Koribalski, B. S. 2001, A&A, 372, 952
Benaglia, P., & Koribalski, B. 2004, A&A, 416, 171
2
See www.aoc.nrao.edu/evla/astro/, www.merlin.ac.uk/e-merlin/.
Non-thermal OB stars
11
Benaglia, P., Romero, G. E., Koribalski, B., & Pollock, A. M. T. 2005, A&A, 440, 743
Benaglia, P., Koribalski, B., & Albacete Colombo, J. F. 2006, PASA, 23, 50
Benaglia, P., & Koribalski, B. 2007, in “Massive Stars in Interacting Binaries”, A. F. J.
Moffat, & N. St-Louis (eds.), ASP Conf. Ser. 367, p. 179
Benaglia, P., Vink, J. S., Martı́, J., et al. 2007, A&A, 467, 1265
Bieging, J. H., Abbott, D. C., & Churchwell, E. B. 1989, ApJ, 340, 518
Biermann, L. 1951, Zs. f. Astrophys. 29, 274
Blomme, R., Prinja, R. K., Runacres, M. C., & Colley, S. 2002, A&A, 382, 921
Blomme, R., van de Steene, G. C., Prinja, R. K., et al. 2003, A&A, 408, 715
Blomme, R., van Loo, S., De Becker, M., et al. 2005, A&A, 436, 1033
Blomme, R., De Becker, M., Runacres, M. C., et al. 2007, A&A, 464, 701
Bulut I., & Demircan O. 2007, MNRAS, 378, 179
Campbell, W. W. 1892, A&A, 11, 799
Castor, J. I., Abbott, D. C., & Klein, R. I. 1975, ApJ, 195, 157
Contreras, M. E., Rodriguez, L. F., Tapia, M., et al. 1997, ApJ, 488, L153
De Becker, M., Rauw, G., Blomme, R., et al. 2004, A&A, 420, 1061
De Becker, M., Rauw, G., Blomme, R., et al. 2005, A&A, 437, 1029
De Becker, M. 2007, ARA&A, 14, 171
Drake, S. A. 1990, AJ, 100, 572
Drake, S. A., Abbott, D. C., Bastian, T. S., et al. 1987, ApJ, 322, 902
Dougherty, S. M. 1993, PhD Thesis Calgary University
Dougherty, S. M., & Kennedy, M. 2009 (in press)
Gibson, D. M., & Hjellming, R. M. 1974, PASP, 88, 652
Howarth, I. D., & Brown, A. 1991, in Proceedings of the IAU Symp. 143, K. A. van der
Hucht & B. Hidayat (eds.), p. 315
Johnson, M. C. 1925, MNRAS, 85, 813
Lamers, H. J. G. L. M., & Leitherer, C. 1993, ApJ, 412, 771
Leitherer, C., & Robert, C. 1991 ApJ, 377, 629
Leitherer, C., Chapman, J. M., & Koribalski, B. 1995, ApJ, 450, 289
Lucy, L. B., & Solomon, P. M. 1970, ApJ, 159, 879
Mason, B. D., Hartkopf, W. I., Gies, D. R., et al. 2009, ApJ, 137, 3358
Maı́z-Apellániz, J., Walborn, N. R., Galué, H. A., & Wei, L. H. 2004, ApJS, 151, 103
Milne, E. A. 1924, MNRAS, 84, 354
Moran, J. P., Davis, R. J., Spencer, R. E., et al. 1989, Nature, 340, 449
Morton, D. C. 1967, ApJ, 147, 1017
Panagia, N., & Felli, M. 1975, A&A, 39, 1
Parker, E. N. 1958, ApJ, 128, 664
Parker, E. N. 1960, ApJ, 132, 821
Persi, P., Rodrı́guez, L. F., Tapia, M., et al. 1988, ASSL, 142, 227
Petr-Gotzens, M. M., & Massi, M. 2008, in “Multiple Stars Across the H-R Diagram”,
Springer-Verlag Berlin Heidelberg, p. 281
Puls, J., Markova, N., Scuderi, S., et al. 2006, A&A, 454, 652
Rauw, G., Blomme, R., Waldron, W. L., et al. 2002, A&A, 394, 993
Romero, G. E., Benaglia, P., & Torres, D. F. 1999, A&A, 348, 868
Schnerr, R. S., Rygl, K. L. J., van der Horst, A. J., et al. 2007, A&A, 470, 1105
Scuderi, S., Panagia, N., Stanghellini, et al. 1998, A&A, 332, 251
Setia Gunawan, D. Y. A., Chapman, J. M., Stevens, I. R., et al. 2003a, Proceedings of
the IAU Symp. 212, K. A. van der Hucht, A. Herrero & C. Esteban (eds.), p. 230
Setia Gunawan, D. Y. A., de Bruyn, A. G., van der Hucht, K. A., & Williams, P. M
2003b, ApJS, 149, 123
Van Loo, S., Blomme, R., Dougherty, S. M., & Runacres, M. C. 2008, A&A, 483, 585
Wendker, H. J. 1995, A&AS, 109, 177
Wolf, C. J. E., & Rayet, G. A. P. 1867, Compte Rendu, 65, 291
Wright, A.E., & Barlow, M.J. 1975, MNRAS, 170, 41